From 5bd044dc61419710ed0430c0fd6d88eedb99eb48 Mon Sep 17 00:00:00 2001 From: lededev Date: Tue, 29 Jun 2021 02:13:46 +0800 Subject: [PATCH] javdb:only accept unique number --- WebCrawler/javdb.py | 13 ++++++------- 1 file changed, 6 insertions(+), 7 deletions(-) mode change 100644 => 100755 WebCrawler/javdb.py diff --git a/WebCrawler/javdb.py b/WebCrawler/javdb.py old mode 100644 new mode 100755 index 97f8f5a..c515255 --- a/WebCrawler/javdb.py +++ b/WebCrawler/javdb.py @@ -216,7 +216,6 @@ def main(number): # else: # number = number.upper() number = number.upper() - isFC2PPV = bool(re.search(r'^FC2-\d+', number)) cookie_json = './' + javdb_site + '.json' javdb_cookies = None # 不加载过期的cookie，javdb登录界面显示为7天免登录，故假定cookie有效期为7天 @@ -241,14 +240,13 @@ def main(number): if re.search(r'[a-zA-Z]+\.\d{2}\.\d{2}\.\d{2}', number): correct_url = urls[0] else: - ids =html.xpath('//*[@id="videos"]/div/div/a/div[contains(@class, "uid")]/text()') + ids = html.xpath('//*[@id="videos"]/div/div/a/div[contains(@class, "uid")]/text()') try: correct_url = urls[ids.index(number)] except: - # 为避免获得错误番号，FC2 PPV 只要精确对应的结果 - if isFC2PPV and ids[0] != number: + # 为避免获得错误番号，只要精确对应的结果 + if ids[0].upper() != number: raise ValueError("number not found") - # if input number is "STAR438" not "STAR-438", use first search result. correct_url = urls[0] try: javdb_detail_url = 'https://' + javdb_site + '.com' + correct_url @@ -273,7 +271,7 @@ def main(number): cover_small = getCover(detail_page) dp_number = getNum(detail_page) - if isFC2PPV and dp_number != number: + if dp_number.upper() != number: raise ValueError("number not found") title = getTitle(detail_page) if title and dp_number: @@ -318,5 +316,6 @@ if __name__ == "__main__": # print(main('AGAV-042')) # print(main('BANK-022')) print(main('FC2-735670')) - print(main('FC2-1174949')) + print(main('FC2-1174949')) # not found print(main('MVSD-439')) + print(main('EHM0001')) # not found